Estimation of Cardiac and Non-cardiac Diagnosis from Electrocardiogram Features

2024-08-30

International Conference on Computing in Cardiology

AI Summary

Electrocardiography (ECG) is traditionally confined to cardiovascular assessment; its potential for predicting non-cardiac diseases remains underexplored. Method: We propose an XGBoost-based multi-label classification framework that jointly models temporal and spectral ECG features with basic demographic information, trained on two multicenter public datasets—MIMIC-IV-ECG-ICD and ECG-VIEW II—to simultaneously predict 23 cardiac and 21 non-cardiac conditions (e.g., diabetes, chronic kidney disease, liver disease). Contribution/Results: This work provides the first systematic evidence that ECG signals robustly support cross-system disease diagnosis: all 44 conditions achieved AUROC > 0.7, with 38 reaching statistical significance (p < 0.05). It challenges the conventional paradigm of ECG use, demonstrating its viability as a non-invasive, low-cost, broad-spectrum screening tool for diverse systemic diseases, thereby establishing both a novel methodological framework and empirical validation for expanded clinical ECG applications.

Technology Category

• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Multiagent Systems: Applications
• Computer Vision: Multi-modal Vision

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Systems and Infrastructure for Web, Mobile and WoT: Web applications in cross-disciplinary domains and verticals such as mixed reality, smart cities, and digital health
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs

Abstract

Introduction: Ensuring timely and accurate diagnosis of medical conditions is paramount for effective patient care. Electrocardiogram (ECG) signals are fundamental for evaluating a patient's cardiac health and are readily available. Despite this, little attention has been given to the remarkable potential of ECG data in detecting non-cardiac conditions. Methods: In our study, we used publicly available datasets (MIMIC-IV-ECG-ICD and ECG-VIEW II) to investigate the feasibility of inferring general diagnostic conditions from ECG features. To this end, we trained a tree-based model (XGBoost) based on ECG features and basic demographic features to estimate a wide range of diagnoses, encompassing both cardiac and non-cardiac conditions. Results: Our results demonstrate the reliability of estimating 23 cardiac as well as 21 non-cardiac conditions above 0.7 AUROC in a statistically significant manner across a wide range of physiological categories. Our findings underscore the predictive potential of ECG data in identifying well-known cardiac conditions. However, even more striking, this research represents a pioneering effort in systematically expanding the scope of ECG-based diagnosis to conditions not traditionally associated with the cardiac system.

Problem

• Estimating both cardiac and non-cardiac diagnoses from ECG features
• Expanding ECG-based diagnosis beyond traditional cardiac conditions
• Investigating ECG data's potential for general diagnostic inference

Innovation

• Used XGBoost model with ECG features
• Estimated cardiac and non-cardiac diagnoses
• Expanded ECG scope to non-traditional conditions